Research map / Reinforcement learning / MDPs and planning
TRUNCATED HORIZON POLICY SEARCH: COMBINING REINFORCEMENT LEARNING & IMITATION LEARNING
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use a planning horizon greater than 1 instead of k=1 (ablation) | THOR with an imperfect oracle · Acrobot | reward | Helped |
| Use k=10 instead of k=1 (ablation) | THOR with an imperfect oracle · Mountain Car | reward | Mixed |
| Use a finite horizon k much smaller than H (ablation) | THOR · Acrobot (H=200) | reward | Helped |
| Use k>1 instead of k=1 (ablation) | THOR with an imperfect value-function oracle · Sparse Reward Inverted Pendulum; Sparse Reward Inverted Double Pendulum; Swimmer; Hopper | reward | Helped |
| Use k=H instead of a truncated horizon (ablation) | THOR with oracle reward shaping · Continuous-action control tasks | reward | Helped |
| Truncate the horizon to reduce policy-gradient estimation variance | THOR · All experiments | variance of performance | Helped |
| Increase the truncation horizon k (ablation) | THOR · Discrete-action control tasks | reward | Helped |
| Use THOR with k>1 rather than AggreVaTeD with k=1 (ablation) | THOR with an imperfect oracle · Sparse-reward discrete-action tasks | reward | Mixed |
Closest papers by meaning
- Beyond-Expert Performance with Limited Demonstrations: Efficient Imitation Learning with Double Exploration (ICLR 2025 · similarity 0.90)
- Blending Imitation and Reinforcement Learning for Robust Policy Improvement (ICLR 2024 · similarity 0.89)
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.88)
- Oracle-Efficient Reinforcement Learning for Max Value Ensembles (NeurIPS 2024 · similarity 0.88)
- Hybrid Inverse Reinforcement Learning (ICML 2024 · similarity 0.88)
- Learning When to Stop: Selective Imitation Learning Under Arbitrary Dynamics Shift (NeurIPS 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.